modeling_gemma4.py

# Copyright 2026 the HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# Adapted for transformers 4.57.1 (trust_remote_code=True usage).

import math
from collections.abc import Callable
from contextlib import contextmanager
from dataclasses import dataclass
from functools import cached_property
from typing import Optional

import torch
import torch.nn.init as init
from torch import nn
from torch.nn import functional as F

from transformers.activations import ACT2FN
from transformers.cache_utils import Cache, DynamicCache
from transformers.configuration_utils import PretrainedConfig
from transformers.generation import GenerationMixin
from transformers.masking_utils import (
    create_causal_mask,
    create_masks_for_generate,
    create_sliding_window_causal_mask,
)
from transformers.modeling_flash_attention_utils import FlashAttentionKwargs
from transformers.modeling_layers import GradientCheckpointingLayer
from transformers.modeling_outputs import BaseModelOutputWithPast, BaseModelOutputWithPooling, CausalLMOutputWithPast
from transformers.modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
from transformers.processing_utils import Unpack
from transformers.utils import ModelOutput, TransformersKwargs, auto_docstring, can_return_tuple

# Local import for trust_remote_code usage
from configuration_gemma4 import Gemma4AudioConfig, Gemma4Config, Gemma4TextConfig, Gemma4VisionConfig


# --- 4.57.1 compat stubs ---

@contextmanager
def maybe_autocast(device_type=None, enabled=True):
    if enabled and device_type is not None:
        with torch.autocast(device_type=device_type):
            yield
    else:
        yield


def merge_with_config_defaults(kwargs, config=None, *keys):
    return kwargs


def use_experts_implementation(cls):
    return cls


def use_kernelized_func(fn):
    def decorator(cls):
        return cls
    return decorator


def torch_compilable_check(condition, message=""):
    if not condition:
        raise ValueError(message)


class OutputRecorder:
    def __init__(self, *args, **kwargs):
        pass


def capture_outputs(fn):
    return fn


def create_bidirectional_mask(attention_mask=None, dtype=None, device=None, config=None, inputs_embeds=None,
                               and_mask_function=None, **kwargs):
    """Bidirectional mask: all-ones (no causal masking), with padding applied."""
    if inputs_embeds is not None and attention_mask is None:
        bsz, seq_len = inputs_embeds.shape[:2]
        attention_mask = torch.ones(bsz, seq_len, device=inputs_embeds.device, dtype=torch.bool)
    elif inputs_embeds is not None:
        bsz, seq_len = inputs_embeds.shape[:2]
    else:
        bsz, seq_len = attention_mask.shape[:2]

    if dtype is None:
        if inputs_embeds is not None:
            dtype = inputs_embeds.dtype
        else:
            dtype = torch.float32
    if device is None:
        if inputs_embeds is not None:
            device = inputs_embeds.device
        elif attention_mask is not None:
            device = attention_mask.device
        else:
            device = torch.device("cpu")

    # Create full bidirectional mask (all zeros = no masking)
    mask = torch.zeros(bsz, 1, seq_len, seq_len, dtype=dtype, device=device)

    # Apply padding mask if attention_mask is a 2D bool/int mask
    if attention_mask is not None and attention_mask.dim() == 2:
        # attention_mask: 1=valid, 0=padding
        pad = (attention_mask == 0)  # [bsz, seq_len]
        # Mask out positions where key is padding
        mask = mask.masked_fill(pad.unsqueeze(1).unsqueeze(2), torch.finfo(dtype).min)

    return mask


def _compute_proportional_rope_parameters(
    config,
    device=None,
    seq_len=None,
    layer_type=None,
    head_dim_key="head_dim",
):
    """
    Proportional RoPE: partial rotary applied to head_dim,
    rest filled with zeros. Ported from transformers 5.x.
    """
    if layer_type is not None:
        rope_params_dict = config.rope_parameters[layer_type]
    else:
        rope_params_dict = config.rope_parameters

    head_dim = getattr(config, head_dim_key, None) or config.hidden_size // config.num_attention_heads

    base = rope_params_dict["rope_theta"]
    factor = rope_params_dict.get("factor", 1.0)
    rope_proportion = rope_params_dict.get("partial_rotary_factor", 1.0)

    attention_factor = 1.0

    rope_angles = int(rope_proportion * head_dim // 2)
    inv_freq_rotated = 1.0 / (
        base
        ** (torch.arange(0, 2 * rope_angles, 2, dtype=torch.int64).to(device=device, dtype=torch.float) / head_dim)
    )

    nope_angles = head_dim // 2 - rope_angles
    if nope_angles > 0:
        inv_freq = torch.cat(
            (inv_freq_rotated, torch.zeros(nope_angles, dtype=torch.float32, device=device)),
            dim=0,
        )
    else:
        inv_freq = inv_freq_rotated

    inv_freq = inv_freq / factor
    return inv_freq, attention_factor


# Patch ROPE_INIT_FUNCTIONS if proportional is missing
if "proportional" not in ROPE_INIT_FUNCTIONS:
    ROPE_INIT_FUNCTIONS["proportional"] = _compute_proportional_rope_parameters


# --- Output dataclasses ---

@dataclass
class Gemma4ModelOutputWithPast(BaseModelOutputWithPast):
    r"""
    Base class for Gemma4 outputs.

    image_hidden_states (`torch.FloatTensor`, *optional*):
        image_hidden_states from the vision encoder.
    audio_hidden_states (`torch.FloatTensor`, *optional*):
        audio_hidden_states from the audio encoder.
    """
    image_hidden_states: Optional[torch.FloatTensor] = None
    audio_hidden_states: Optional[torch.FloatTensor] = None


@dataclass
class Gemma4CausalLMOutputWithPast(ModelOutput):
    r"""
    Causal LM output for Gemma4.
    """
    loss: Optional[torch.FloatTensor] = None
    logits: Optional[torch.FloatTensor] = None
    past_key_values: Optional[Cache] = None
    hidden_states: Optional[tuple] = None
    attentions: Optional[tuple] = None
    image_hidden_states: Optional[torch.FloatTensor] = None
    audio_hidden_states: Optional[torch.FloatTensor] = None


@dataclass
class Gemma4AudioModelOutput(BaseModelOutputWithPooling):
    r"""
    Audio model output.

    attention_mask (`torch.BoolTensor`, *optional*):
        True for valid positions, False for padding.
    """
    attention_mask: Optional[torch.BoolTensor] = None


# --- Modules ---

class Gemma4ClippableLinear(nn.Module):
    def __init__(
        self,
        config,
        in_features: int,
        out_features: int,
    ) -> None:
        super().__init__()
        self.use_clipped_linears = config.use_clipped_linears
        self.linear = nn.Linear(in_features, out_features, bias=False)

        if self.use_clipped_linears:
            self.register_buffer("input_min", torch.tensor(-float("inf")))
            self.register_buffer("input_max", torch.tensor(float("inf")))
            self.register_buffer("output_min", torch.tensor(-float("inf")))
            self.register_buffer("output_max", torch.tensor(float("inf")))

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        if self.use_clipped_linears:
            hidden_states = torch.clamp(hidden_states, self.input_min, self.input_max)
        hidden_states = self.linear(hidden_states)
        if self.use_clipped_linears:
            hidden_states = torch.clamp(hidden_states, self.output_min, self.output_max)
        return hidden_states


class Gemma4RMSNorm(nn.Module):
    def __init__(self, dim: int, eps: float = 1e-6, with_scale: bool = True):
        super().__init__()
        self.eps = eps
        self.with_scale = with_scale

        if self.with_scale:
            self.weight = nn.Parameter(torch.ones(dim), requires_grad=True)

    def _norm(self, hidden_states: torch.Tensor):
        mean_squared = hidden_states.pow(2).mean(-1, keepdim=True) + self.eps
        return hidden_states * torch.pow(mean_squared, -0.5)

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        normed_output = self._norm(hidden_states.float())
        if self.with_scale:
            normed_output = normed_output * self.weight.float()
        return normed_output.type_as(hidden_states)


class Gemma4AudioRelPositionalEncoding(nn.Module):
    """Sinusoidal relative positional encoding for the audio encoder."""

    inv_timescales: torch.Tensor

    def __init__(self, config: Gemma4AudioConfig):
        super().__init__()
        self.hidden_size = config.hidden_size
        self.context_size = (
            config.attention_chunk_size + config.attention_context_left - 1 + config.attention_context_right
        )
        min_timescale = 1.0
        max_timescale = 10000.0
        num_timescales = self.hidden_size // 2
        log_timescale_increment = math.log(max_timescale / min_timescale) / max(num_timescales - 1, 1)
        inv_timescales = min_timescale * torch.exp(torch.arange(num_timescales) * -log_timescale_increment)
        self.register_buffer("inv_timescales", inv_timescales.unsqueeze(0).unsqueeze(0), persistent=False)

    @torch.no_grad()
    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        position_ids = torch.arange(12, -1, -1, device=hidden_states.device)
        position_ids = position_ids[..., None]
        scaled_time = position_ids * self.inv_timescales.to(device=hidden_states.device)
        pos_embed = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], dim=-1)
        return pos_embed.to(dtype=hidden_states.dtype)


class Gemma4AudioAttention(nn.Module):
    """Chunked local attention with relative position bias"""

    def __init__(self, config: Gemma4AudioConfig, layer_idx: int):
        super().__init__()
        self.config = config
        self.layer_idx = layer_idx
        self.attention_logits_soft_cap = config.attention_logit_cap
        self.head_dim = config.hidden_size // config.num_attention_heads
        self.num_heads = config.num_attention_heads

        self.q_scale = (self.head_dim**-0.5) / math.log(2)
        self.k_scale = math.log(1 + math.e) / math.log(2)

        self.chunk_size = config.attention_chunk_size
        self.max_past_horizon = config.attention_context_left - 1
        self.max_future_horizon = config.attention_context_right
        self.context_size = self.chunk_size + self.max_past_horizon + self.max_future_horizon

        self.q_proj = Gemma4ClippableLinear(config, config.hidden_size, self.num_heads * self.head_dim)
        self.k_proj = Gemma4ClippableLinear(config, config.hidden_size, self.num_heads * self.head_dim)
        self.v_proj = Gemma4ClippableLinear(config, config.hidden_size, self.num_heads * self.head_dim)
        self.post = Gemma4ClippableLinear(config, config.hidden_size, config.hidden_size)

        self.relative_k_proj = nn.Linear(config.hidden_size, self.num_heads * self.head_dim, bias=False)
        self.per_dim_scale = nn.Parameter(torch.zeros(self.head_dim))

        self.register_buffer("softcap", torch.tensor(self.attention_logits_soft_cap), persistent=False)

    def _convert_to_block(self, hidden_states: torch.Tensor) -> torch.Tensor:
        batch_size, seq_len, num_heads, head_dim = hidden_states.shape
        num_blocks = (seq_len + self.chunk_size - 1) // self.chunk_size
        pad = num_blocks * self.chunk_size - seq_len
        hidden_states = F.pad(hidden_states, (0, 0, 0, 0, 0, pad))
        return hidden_states.reshape(batch_size, num_blocks, self.chunk_size, num_heads, head_dim).contiguous()

    def _extract_block_context(self, hidden_states: torch.Tensor) -> torch.Tensor:
        batch_size, seq_len, num_heads, head_dim = hidden_states.shape
        hidden_states = F.pad(
            hidden_states, (0, 0, 0, 0, self.max_past_horizon, self.max_future_horizon + self.chunk_size - 1)
        )
        hidden_states = hidden_states.unfold(1, self.context_size, self.chunk_size)
        hidden_states = torch.movedim(hidden_states, -1, 2)
        return hidden_states.contiguous()

    def _rel_shift(self, x: torch.Tensor) -> torch.Tensor:
        batch_size, num_heads, num_blocks, block_size, position_length = x.shape
        context_size = self.context_size
        x = F.pad(x, (0, context_size + 1 - position_length))
        x = x.view(batch_size, num_heads, num_blocks, block_size * (context_size + 1))
        x = x[..., : block_size * context_size]
        return x.view(batch_size, num_heads, num_blocks, block_size, context_size)

    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings: torch.Tensor,
        attention_mask=None,
    ):
        batch_size, seq_length, _ = hidden_states.shape
        hidden_shape = (batch_size, seq_length, self.num_heads, self.head_dim)

        query_states = self.q_proj(hidden_states).float().view(hidden_shape)
        key_states = self.k_proj(hidden_states).float().view(hidden_shape)
        value_states = self.v_proj(hidden_states).float().view(hidden_shape)

        query_states = query_states * self.q_scale * F.softplus(self.per_dim_scale)
        key_states = key_states * self.k_scale

        query_states = self._convert_to_block(query_states)
        key_states = self._extract_block_context(key_states)
        value_states = self._extract_block_context(value_states)
        num_blocks = query_states.shape[1]

        relative_key_states = self.relative_k_proj(position_embeddings)
        relative_key_states = relative_key_states.view(-1, self.num_heads, self.head_dim)
        relative_key_states = relative_key_states.to(dtype=query_states.dtype)

        queries = query_states.permute(0, 3, 1, 2, 4)
        matrix_ac = queries @ key_states.permute(0, 3, 1, 4, 2)

        queries_flat = queries.reshape(batch_size, self.num_heads, -1, self.head_dim)
        matrix_bd = queries_flat @ relative_key_states.permute(1, 2, 0)
        matrix_bd = matrix_bd.reshape(batch_size, self.num_heads, num_blocks, self.chunk_size, -1)
        matrix_bd = self._rel_shift(matrix_bd)

        attn_weights = matrix_ac + matrix_bd
        attn_weights = attn_weights / self.softcap
        attn_weights = torch.tanh(attn_weights)
        attn_weights = attn_weights * self.softcap

        if attention_mask is not None:
            attn_weights = attn_weights.masked_fill(
                attention_mask.logical_not(), self.config.attention_invalid_logits_value
            )

        attn_weights = F.softmax(attn_weights, dim=-1, dtype=torch.float32).to(value_states.dtype)
        attn_output = attn_weights @ value_states.permute(0, 3, 1, 2, 4)
        attn_output = attn_output.permute(0, 2, 3, 1, 4).reshape(batch_size, num_blocks * self.chunk_size, -1)
        attn_output = attn_output[:, :seq_length].contiguous()
        attn_output = self.post(attn_output.to(dtype=self.post.linear.weight.dtype))

        return attn_output, attn_weights


class Gemma4AudioSubSampleConvProjectionLayer(nn.Module):
    def __init__(self, in_channels, out_channels, norm_eps):
        super().__init__()
        self.conv = nn.Conv2d(
            in_channels=in_channels,
            out_channels=out_channels,
            kernel_size=(3, 3),
            stride=(2, 2),
            padding=1,
            bias=False,
        )
        self.norm = nn.LayerNorm(out_channels, eps=norm_eps, elementwise_affine=True, bias=False)
        self.act = nn.ReLU()

    def forward(self, hidden_states: torch.Tensor, mask=None):
        if mask is not None:
            mask = mask.to(device=hidden_states.device)
            hidden_states = hidden_states * mask[:, None, :, None]

        hidden_states = self.conv(hidden_states.to(self.conv.weight.dtype))
        hidden_states = self.act(self.norm(hidden_states.permute(0, 2, 3, 1)).permute(0, 3, 1, 2).contiguous())

        if mask is not None:
            mask = mask[:, ::2]

        return hidden_states, mask


class Gemma4AudioSubSampleConvProjection(nn.Module):
    def __init__(self, config: Gemma4AudioConfig):
        super().__init__()
        self.layer0 = Gemma4AudioSubSampleConvProjectionLayer(
            in_channels=1,
            out_channels=config.subsampling_conv_channels[0],
            norm_eps=config.rms_norm_eps,
        )
        self.layer1 = Gemma4AudioSubSampleConvProjectionLayer(
            in_channels=config.subsampling_conv_channels[0],
            out_channels=config.subsampling_conv_channels[1],
            norm_eps=config.rms_norm_eps,
        )
        proj_input_dim = (config.subsampling_conv_channels[0] // 4) * config.subsampling_conv_channels[1]
        self.input_proj_linear = nn.Linear(proj_input_dim, config.hidden_size, bias=False)

    def forward(self, input_features: torch.Tensor, input_features_mask=None):
        hidden_states = input_features.unsqueeze(1)
        hidden_states, mask = self.layer0(hidden_states, input_features_mask)
        hidden_states, mask = self.layer1(hidden_states, mask)

        batch_size, _, seq_len, _ = hidden_states.shape
        hidden_states = hidden_states.permute(0, 2, 3, 1).contiguous().reshape(batch_size, seq_len, -1)
        return self.input_proj_linear(hidden_states), mask


class Gemma4AudioFeedForward(nn.Module):
    def __init__(self, config: Gemma4AudioConfig):
        super().__init__()
        self.config = config

        self.ffw_layer_1 = Gemma4ClippableLinear(config, config.hidden_size, config.hidden_size * 4)
        self.ffw_layer_2 = Gemma4ClippableLinear(config, config.hidden_size * 4, config.hidden_size)

        self.pre_layer_norm = Gemma4RMSNorm(config.hidden_size)
        self.post_layer_norm = Gemma4RMSNorm(config.hidden_size)
        self.act_fn = ACT2FN[config.hidden_act]

        self.gradient_clipping = config.gradient_clipping
        self.post_layer_scale = config.residual_weight

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        gradient_clipping = min(self.gradient_clipping, torch.finfo(self.ffw_layer_1.linear.weight.dtype).max)

        residual = hidden_states
        hidden_states = torch.clamp(hidden_states, -gradient_clipping, gradient_clipping)
        hidden_states = self.pre_layer_norm(hidden_states)

        hidden_states = self.ffw_layer_1(hidden_states)
        hidden_states = self.act_fn(hidden_states)
        hidden_states = self.ffw_layer_2(hidden_states)

        hidden_states = torch.clamp(hidden_states, -gradient_clipping, gradient_clipping)
        hidden_states = self.post_layer_norm(hidden_states)
        hidden_states *= self.post_layer_scale
        hidden_states += residual

        return hidden_states


class Gemma4AudioCausalConv1d(nn.Conv1d):
    @cached_property
    def left_pad(self):
        effective_kernel_size = (self.kernel_size[0] - 1) * self.dilation[0] + 1
        return effective_kernel_size - self.stride[0]

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        x = nn.functional.pad(x, (self.left_pad, 0))
        return super().forward(x)


class Gemma4AudioLightConv1d(nn.Module):
    def __init__(self, config: Gemma4AudioConfig):
        super().__init__()
        self.config = config

        self.linear_start = Gemma4ClippableLinear(config, config.hidden_size, config.hidden_size * 2)
        self.linear_end = Gemma4ClippableLinear(config, config.hidden_size, config.hidden_size)
        self.depthwise_conv1d = Gemma4AudioCausalConv1d(
            in_channels=config.hidden_size,
            out_channels=config.hidden_size,
            kernel_size=config.conv_kernel_size,
            groups=config.hidden_size,
            bias=False,
        )

        self.pre_layer_norm = Gemma4RMSNorm(config.hidden_size, eps=config.rms_norm_eps, with_scale=True)
        self.conv_norm = Gemma4RMSNorm(config.hidden_size, eps=config.rms_norm_eps, with_scale=True)
        self.act_fn = ACT2FN[config.hidden_act]

        self.gradient_clipping = config.gradient_clipping

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        residual = hidden_states

        hidden_states = self.pre_layer_norm(hidden_states)
        hidden_states = self.linear_start(hidden_states)
        hidden_states = nn.functional.glu(hidden_states, dim=-1)

        hidden_states = self.depthwise_conv1d(hidden_states.transpose(1, 2)).transpose(1, 2)

        gradient_clipping = min(self.gradient_clipping, torch.finfo(self.linear_start.linear.weight.dtype).max)
        hidden_states = torch.clamp(hidden_states, -gradient_clipping, gradient_clipping)
        hidden_states = self.conv_norm(hidden_states)

        hidden_states = self.act_fn(hidden_states)
        hidden_states = self.linear_end(hidden_states)
        hidden_states += residual
        return hidden_states


class Gemma4AudioLayer(nn.Module):
    def __init__(self, config: Gemma4AudioConfig, layer_idx: int):
        super().__init__()
        self.config = config

        self.feed_forward1 = Gemma4AudioFeedForward(config)
        self.feed_forward2 = Gemma4AudioFeedForward(config)
        self.self_attn = Gemma4AudioAttention(config, layer_idx)
        self.lconv1d = Gemma4AudioLightConv1d(config)

        self.norm_pre_attn = Gemma4RMSNorm(config.hidden_size)
        self.norm_post_attn = Gemma4RMSNorm(config.hidden_size)
        self.norm_out = Gemma4RMSNorm(config.hidden_size)

        self.gradient_clipping = config.gradient_clipping

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask=None,
        position_embeddings: torch.Tensor = None,
        **kwargs,
    ) -> torch.Tensor:
        gradient_clipping = min(self.gradient_clipping, torch.finfo(self.norm_pre_attn.weight.dtype).max)

        hidden_states = self.feed_forward1(hidden_states)
        residual = hidden_states

        hidden_states = torch.clamp(hidden_states, -gradient_clipping, gradient_clipping)
        hidden_states = self.norm_pre_attn(hidden_states)

        hidden_states, _ = self.self_attn(
            hidden_states=hidden_states,
            position_embeddings=position_embeddings,
            attention_mask=attention_mask,
        )

        hidden_states = torch.clamp(hidden_states, -gradient_clipping, gradient_clipping)
        hidden_states = self.norm_post_attn(hidden_states)
        hidden_states += residual

        hidden_states = self.lconv1d(hidden_states)
        hidden_states = self.feed_forward2(hidden_states)

        hidden_states = torch.clamp(hidden_states, -gradient_clipping, gradient_clipping)
        hidden_states = self.norm_out(hidden_states)

        return hidden_states


# ---- Vision Encoder Layers ----

class Gemma4VisionPatchEmbedder(nn.Module):
    def __init__(self, config: Gemma4VisionConfig):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.patch_size = config.patch_size
        self.position_embedding_size = config.position_embedding_size

        self.input_proj = nn.Linear(3 * self.patch_size**2, self.hidden_size, bias=False)
        self.position_embedding_table = nn.Parameter(torch.ones(2, self.position_embedding_size, self.hidden_size))

    def _position_embeddings(self, pixel_position_ids: torch.Tensor, padding_positions: torch.Tensor) -> torch.Tensor:
        clamped_positions = pixel_position_ids.clamp(min=0)
        one_hot = F.one_hot(clamped_positions, num_classes=self.position_embedding_size)
        one_hot = one_hot.permute(0, 2, 1, 3).to(self.position_embedding_table)
        position_embeddings = one_hot @ self.position_embedding_table
        position_embeddings = position_embeddings.sum(dim=1)
        position_embeddings = torch.where(padding_positions.unsqueeze(-1), 0.0, position_embeddings)
        return position_embeddings

    def forward(
        self, pixel_values: torch.Tensor, pixel_position_ids: torch.Tensor, padding_positions: torch.Tensor
    ) -> torch.Tensor:
        pixel_values = 2 * (pixel_values - 0.5)
        hidden_states = self.input_proj(pixel_values.to(self.input_proj.weight.dtype))
        position_embeddings = self._position_embeddings(pixel_position_ids, padding_positions)
        return hidden_states + position_embeddings


class Gemma4VisionPooler(nn.Module):
    """Scaling and optional spatial pooling for vision encodings"""

    def __init__(self, config: Gemma4VisionConfig):
        super().__init__()
        self.hidden_size = config.hidden_size
        self.root_hidden_size = self.hidden_size**0.5

    def _avg_pool_by_positions(self, hidden_states, pixel_position_ids, length):
        input_seq_len = hidden_states.shape[1]
        k = int((input_seq_len // length) ** 0.5)
        k_squared = k**2
        if k_squared * length != input_seq_len:
            raise ValueError(
                f"Cannot pool {hidden_states.shape} to {length}: {k=}^2 times {length=} must be {input_seq_len}."
            )

        clamped_positions = pixel_position_ids.clamp(min=0)
        max_x = clamped_positions[..., 0].max(dim=-1, keepdim=True)[0] + 1
        kernel_idxs = torch.div(clamped_positions, k, rounding_mode="floor")
        kernel_idxs = kernel_idxs[..., 0] + (max_x // k) * kernel_idxs[..., 1]
        weights = F.one_hot(kernel_idxs.long(), length).float() / k_squared
        output = weights.transpose(1, 2) @ hidden_states.float()
        mask = torch.logical_not((weights == 0).all(dim=1))
        return output.to(hidden_states.dtype), mask

    def forward(self, hidden_states, pixel_position_ids, padding_positions, output_length=None):
        if output_length > hidden_states.shape[1]:
            raise ValueError(
                f"Cannot output more soft tokens (requested {output_length}) than there are patches"
                f" ({hidden_states.shape[1]}). Change the value of `num_soft_tokens` when processing."
            )

        hidden_states = hidden_states.masked_fill(padding_positions.unsqueeze(-1), 0.0)

        if hidden_states.shape[1] != output_length:
            hidden_states, padding_positions = self._avg_pool_by_positions(
                hidden_states, pixel_position_ids, output_length
            )

        hidden_states *= self.root_hidden_size
        return hidden_states, padding_positions


class Gemma4VisionMLP(nn.Module):
    def __init__(self, config: Gemma4VisionConfig):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.intermediate_size = config.intermediate_size
        self.gate_proj = Gemma4ClippableLinear(config, self.hidden_size, self.intermediate_size)
        self.up_proj = Gemma4ClippableLinear(config, self.hidden_size, self.intermediate_size)
        self.down_proj = Gemma4ClippableLinear(config, self.intermediate_size, self.hidden_size)
        self.act_fn = ACT2FN[config.hidden_activation]

    def forward(self, x):
        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))


class Gemma4VisionRotaryEmbedding(nn.Module):
    inv_freq: torch.Tensor

    def __init__(self, config: Gemma4VisionConfig, device=None):
        super().__init__()
        self.max_seq_len_cached = config.max_position_embeddings
        self.original_max_seq_len = config.max_position_embeddings

        self.config = config

        self.rope_type = self.config.rope_parameters["rope_type"]
        if self.rope_type != "default":
            rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
        else:
            rope_init_fn = self.compute_default_rope_parameters
        inv_freq, self.attention_scaling = rope_init_fn(self.config, device)

        self.register_buffer("inv_freq", inv_freq, persistent=False)
        self.register_buffer("original_inv_freq", inv_freq.clone(), persistent=False)

    @staticmethod
    def compute_default_rope_parameters(config=None, device=None, seq_len=None):
        base = config.rope_parameters["rope_theta"]
        dim = getattr(config, "head_dim", None) or config.hidden_size // config.num_attention_heads

        spatial_dim = dim // 2
        attention_factor = 1.0
        inv_freq = 1.0 / (
            base
            ** (torch.arange(0, spatial_dim, 2, dtype=torch.int64).to(device=device, dtype=torch.float) / spatial_dim)
        )
        return inv_freq, attention_factor

    @torch.no_grad()
    @dynamic_rope_update
    def forward(self, x, position_ids):
        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"

        all_cos, all_sin = [], []
        for i in range(2):
            dim_position_ids = position_ids[:, :, i]
            dim_position_ids_expanded = dim_position_ids[:, None, :].float()

            with maybe_autocast(device_type=device_type, enabled=False):
                freqs = (inv_freq_expanded.float() @ dim_position_ids_expanded.float()).transpose(1, 2)
                emb = torch.cat((freqs, freqs), dim=-1)
                cos = emb.cos() * self.attention_scaling
                sin = emb.sin() * self.attention_scaling
            all_cos.append(cos)
            all_sin.append(sin)

        cos = torch.cat(all_cos, dim=-1).to(dtype=x.dtype)
        sin = torch.cat(all_sin, dim=-1).to(dtype=x.dtype)
        return cos, sin


def rotate_half(x):
    """Rotates half the hidden dims of the input."""
    x1 = x[..., : x.shape[-1] // 2]
    x2 = x[..., x.shape[-1] // 2:]
    return torch.cat((-x2, x1), dim=-1)


def apply_rotary_pos_emb(x: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
    cos = cos.unsqueeze(unsqueeze_dim)
    sin = sin.unsqueeze(unsqueeze_dim)
    return (x * cos) + (rotate_half(x) * sin)


def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
    if n_rep == 1:
        return hidden_states
    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)


def eager_attention_forward(
    module: nn.Module,
    query: torch.Tensor,
    key: torch.Tensor,
    value: torch.Tensor,
    attention_mask,
    dropout: float = 0.0,
    scaling=None,
    softcap=None,
    **kwargs,
):
    if scaling is None:
        scaling = module.head_dim**-0.5

    key_states = repeat_kv(key, module.num_key_value_groups)
    value_states = repeat_kv(value, module.num_key_value_groups)

    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling

    if softcap is not None:
        attn_weights = attn_weights / softcap
        attn_weights = torch.tanh(attn_weights)
        attn_weights = attn_weights * softcap
    if attention_mask is not None:
        attn_weights = attn_weights + attention_mask

    attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
    attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)
    attn_output = torch.matmul(attn_weights, value_states)
    attn_output = attn_output.transpose(1, 2).contiguous()
    return attn_output, attn_weights


def apply_multidimensional_rope(x, cos, sin, position_ids, unsqueeze_dim=2):
    ndim = position_ids.shape[-1]
    num_input_channels = x.shape[-1]
    num_rotated_channels_per_dim = 2 * (num_input_channels // (2 * ndim))

    if num_rotated_channels_per_dim <= 0:
        raise ValueError(
            f"Invalid configuration: num_rotated_channels_per_dim must be > 0, got"
            f" {num_rotated_channels_per_dim} (num_input_channels={num_input_channels}, ndim={ndim})"
        )

    split_sizes = [num_rotated_channels_per_dim] * ndim
    x_parts = torch.split(x, split_sizes, dim=-1)
    cos_parts = torch.split(cos, split_sizes, dim=-1)
    sin_parts = torch.split(sin, split_sizes, dim=-1)
    y_parts = [
        apply_rotary_pos_emb(x=x_parts[k], cos=cos_parts[k], sin=sin_parts[k], unsqueeze_dim=unsqueeze_dim)
        for k in range(ndim)
    ]
    return torch.cat(y_parts, dim=-1)


@use_kernelized_func(apply_rotary_pos_emb)
class Gemma4VisionAttention(nn.Module):
    """Multi-headed attention from 'Attention Is All You Need' paper"""

    def __init__(self, config: Gemma4VisionConfig, layer_idx: int):
        super().__init__()
        self.layer_type = config.layer_types[layer_idx] if hasattr(config, "layer_types") else None
        self.config = config
        self.layer_idx = layer_idx
        self.head_dim = getattr(config, "head_dim", config.hidden_size // config.num_attention_heads)
        self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
        self.scaling = 1.0
        self.attention_dropout = self.config.attention_dropout
        self.is_causal = False
        self.q_proj = Gemma4ClippableLinear(config, config.hidden_size, config.num_attention_heads * self.head_dim)
        self.k_proj = Gemma4ClippableLinear(config, config.hidden_size, config.num_key_value_heads * self.head_dim)
        self.v_proj = Gemma4ClippableLinear(config, config.hidden_size, config.num_key_value_heads * self.head_dim)
        self.o_proj = Gemma4ClippableLinear(config, config.num_attention_heads * self.head_dim, config.hidden_size)

        self.q_norm = Gemma4RMSNorm(dim=config.head_dim, eps=config.rms_norm_eps)
        self.k_norm = Gemma4RMSNorm(dim=config.head_dim, eps=config.rms_norm_eps)
        self.v_norm = Gemma4RMSNorm(self.head_dim, eps=config.rms_norm_eps, with_scale=False)

    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings=None,
        attention_mask=None,
        position_ids=None,
        **kwargs,
    ):
        input_shape = hidden_states.shape[:-1]
        hidden_shape = (*input_shape, -1, self.head_dim)

        cos, sin = position_embeddings

        query_states = self.q_proj(hidden_states).view(hidden_shape)
        query_states = self.q_norm(query_states)
        query_states = apply_multidimensional_rope(query_states, cos, sin, position_ids)
        query_states = query_states.transpose(1, 2)

        key_states = self.k_proj(hidden_states).view(hidden_shape)
        key_states = self.k_norm(key_states)
        key_states = apply_multidimensional_rope(key_states, cos, sin, position_ids)
        key_states = key_states.transpose(1, 2)

        value_states = self.v_proj(hidden_states).view(hidden_shape)
        value_states = self.v_norm(value_states)
        value_states = value_states.transpose(1, 2)

        # In 4.57.1, ALL_ATTENTION_FUNCTIONS is a dict-like, use .get with fallback
        attn_impl = self.config._attn_implementation
        if attn_impl != "eager" and attn_impl in ALL_ATTENTION_FUNCTIONS:
            attention_interface = ALL_ATTENTION_FUNCTIONS[attn_impl]
        else:
            attention_interface = eager_attention_forward

        attn_output, attn_weights = attention_interface(
            self,
            query_states,
            key_states,
            value_states,
            attention_mask,
            dropout=self.attention_dropout if self.training else 0.0,
            scaling=self.scaling,
            **kwargs,
        )

        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
        attn_output = self.o_proj(attn_output)
        return attn_output, attn_weights


class Gemma4VisionEncoderLayer(GradientCheckpointingLayer):
    def __init__(self, config: Gemma4VisionConfig, layer_idx: int):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.layer_idx = layer_idx
        self.self_attn = Gemma4VisionAttention(config=config, layer_idx=layer_idx)
        self.mlp = Gemma4VisionMLP(config)
        self.input_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
        self.post_attention_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
        self.pre_feedforward_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
        self.post_feedforward_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)

    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings=None,
        attention_mask=None,
        position_ids=None,
        **kwargs,
    ):
        residual = hidden_states

        hidden_states = self.input_layernorm(hidden_states)
        hidden_states, _ = self.self_attn(
            hidden_states=hidden_states,
            position_embeddings=position_embeddings,
            attention_mask=attention_mask,
            position_ids=position_ids,
            **kwargs,
        )
        hidden_states = self.post_attention_layernorm(hidden_states)
        hidden_states = residual + hidden_states

        residual = hidden_states
        hidden_states = self.pre_feedforward_layernorm(hidden_states)
        hidden_states = self.mlp(hidden_states)
        hidden_states = self.post_feedforward_layernorm(hidden_states)
        hidden_states = residual + hidden_states

        return hidden_states


class Gemma4VisionEncoder(nn.Module):
    def __init__(self, config: Gemma4VisionConfig):
        super().__init__()
        self.config = config
        self.num_layers = config.num_hidden_layers
        self.rotary_emb = Gemma4VisionRotaryEmbedding(config)
        self.layers = nn.ModuleList(
            [Gemma4VisionEncoderLayer(config=config, layer_idx=i) for i in range(self.num_layers)]
        )

    def forward(self, inputs_embeds, attention_mask, pixel_position_ids=None, **kwargs):
        attention_mask = create_bidirectional_mask(
            config=self.config,
            inputs_embeds=inputs_embeds,
            attention_mask=attention_mask,
        )

        hidden_states = inputs_embeds
        position_embeddings = self.rotary_emb(hidden_states, pixel_position_ids)

        for decoder_layer in self.layers[: self.config.num_hidden_layers]:
            hidden_states = decoder_layer(
                hidden_states,
                attention_mask=attention_mask,
                position_embeddings=position_embeddings,
                position_ids=pixel_position_ids,
                **kwargs,
            )

        return BaseModelOutputWithPast(last_hidden_state=hidden_states)


class Gemma4TextMLP(nn.Module):
    def __init__(self, config: Gemma4TextConfig, layer_idx: int):
        super().__init__()
        first_kv_shared_layer_idx = config.num_hidden_layers - config.num_kv_shared_layers
        is_kv_shared_layer = layer_idx >= first_kv_shared_layer_idx > 0
        use_double_wide_mlp = config.use_double_wide_mlp and is_kv_shared_layer
        self.config = config
        self.hidden_size = config.hidden_size
        self.intermediate_size = config.intermediate_size * (2 if use_double_wide_mlp else 1)
        self.gate_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
        self.up_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
        self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=False)
        self.act_fn = ACT2FN[config.hidden_activation]

    def forward(self, x):
        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))


class Gemma4TextRotaryEmbedding(nn.Module):
    inv_freq: torch.Tensor

    def __init__(self, config: Gemma4TextConfig, device=None, layer_type=None):
        super().__init__()
        self.max_seq_len_cached = config.max_position_embeddings
        self.original_max_seq_len = config.max_position_embeddings

        self.config = config
        self.layer_types = set(config.layer_types)
        self.rope_init_fns = {}
        self.rope_type = {}

        for lt in self.layer_types:
            rope_params = self.config.rope_parameters[lt]
            if rope_params is None:
                continue

            rope_type_name = rope_params["rope_type"]
            if rope_type_name != "default":
                rope_init_fn = ROPE_INIT_FUNCTIONS[rope_type_name]
            else:
                rope_init_fn = self.compute_default_rope_parameters

            self.rope_init_fns[lt] = rope_init_fn
            self.rope_type[lt] = rope_type_name

            rope_init_fn_kwargs = {"device": device, "layer_type": lt}
            if lt == "full_attention" and rope_type_name == "proportional":
                rope_init_fn_kwargs["head_dim_key"] = "global_head_dim"

            curr_inv_freq, curr_attention_scaling = rope_init_fn(self.config, **rope_init_fn_kwargs)
            self.register_buffer(f"{lt}_inv_freq", curr_inv_freq, persistent=False)
            self.register_buffer(f"{lt}_original_inv_freq", curr_inv_freq.clone(), persistent=False)
            setattr(self, f"{lt}_attention_scaling", curr_attention_scaling)

    @staticmethod
    def compute_default_rope_parameters(config=None, device=None, seq_len=None, layer_type=None, **kwargs):
        base = config.rope_parameters[layer_type]["rope_theta"]
        dim = getattr(config, "head_dim", None) or config.hidden_size // config.num_attention_heads
        attention_factor = 1.0
        inv_freq = 1.0 / (
            base ** (torch.arange(0, dim, 2, dtype=torch.int64).to(device=device, dtype=torch.float) / dim)
        )
        return inv_freq, attention_factor

    @torch.no_grad()
    @dynamic_rope_update
    def forward(self, x, position_ids, layer_type=None):
        inv_freq = getattr(self, f"{layer_type}_inv_freq")
        attention_scaling = getattr(self, f"{layer_type}_attention_scaling")

        inv_freq_expanded = inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
        position_ids_expanded = position_ids[:, None, :].float()

        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
        with maybe_autocast(device_type=device_type, enabled=False):
            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
            emb = torch.cat((freqs, freqs), dim=-1)
            cos = emb.cos() * attention_scaling
            sin = emb.sin() * attention_scaling

        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)


@use_kernelized_func(apply_rotary_pos_emb)
class Gemma4TextAttention(nn.Module):
    """Multi-headed attention from 'Attention Is All You Need' paper"""

    def __init__(self, config: Gemma4TextConfig, layer_idx: int):
        super().__init__()
        self.layer_type = config.layer_types[layer_idx] if hasattr(config, "layer_types") else None
        self.config = config
        self.layer_idx = layer_idx
        self.is_sliding = self.layer_type == "sliding_attention"
        self.sliding_window = config.sliding_window if self.is_sliding else None

        self.head_dim = config.global_head_dim if not self.is_sliding and config.global_head_dim else config.head_dim
        self.use_alternative_attention = config.attention_k_eq_v and not self.is_sliding
        num_key_value_heads = (
            config.num_global_key_value_heads if self.use_alternative_attention else config.num_key_value_heads
        )
        self.num_key_value_groups = config.num_attention_heads // num_key_value_heads
        self.scaling = 1.0
        self.attention_dropout = self.config.attention_dropout
        self.is_causal = config.use_bidirectional_attention != "all"

        # Shared kv cache
        first_kv_shared_layer_idx = self.config.num_hidden_layers - getattr(self.config, "num_kv_shared_layers", 0)
        self.is_kv_shared_layer = layer_idx >= first_kv_shared_layer_idx > 0
        prev_layers = config.layer_types[:first_kv_shared_layer_idx]
        if self.is_kv_shared_layer:
            self.kv_shared_layer_index = len(prev_layers) - 1 - prev_layers[::-1].index(config.layer_types[layer_idx])
            self.store_full_length_kv = False
        else:
            self.kv_shared_layer_index = None
            self.store_full_length_kv = layer_idx == len(prev_layers) - 1 - prev_layers[::-1].index(
                config.layer_types[layer_idx]
            )

        self.q_proj = nn.Linear(
            config.hidden_size, config.num_attention_heads * self.head_dim, bias=config.attention_bias
        )
        self.q_norm = Gemma4RMSNorm(dim=self.head_dim, eps=config.rms_norm_eps)

        if not self.is_kv_shared_layer:
            self.k_norm = Gemma4RMSNorm(dim=self.head_dim, eps=config.rms_norm_eps)
            self.v_norm = Gemma4RMSNorm(self.head_dim, eps=config.rms_norm_eps, with_scale=False)

            self.k_proj = nn.Linear(
                config.hidden_size, num_key_value_heads * self.head_dim, bias=config.attention_bias
            )
            self.v_proj = (
                nn.Linear(config.hidden_size, num_key_value_heads * self.head_dim, bias=config.attention_bias)
                if not self.use_alternative_attention
                else None
            )

        self.o_proj = nn.Linear(
            config.num_attention_heads * self.head_dim, config.hidden_size, bias=config.attention_bias
        )

    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings: torch.Tensor,
        attention_mask,
        shared_kv_states,
        past_key_values=None,
        **kwargs,
    ):
        input_shape = hidden_states.shape[:-1]
        hidden_shape = (*input_shape, -1, self.head_dim)

        cos, sin = position_embeddings

        query_states = self.q_proj(hidden_states).view(hidden_shape)
        query_states = self.q_norm(query_states)
        query_states = apply_rotary_pos_emb(query_states, cos, sin, unsqueeze_dim=2)
        query_states = query_states.transpose(1, 2)

        if self.is_kv_shared_layer:
            key_states, value_states = shared_kv_states[self.kv_shared_layer_index]
            key_states = key_states.to(query_states.device)
            value_states = value_states.to(query_states.device)
        else:
            key_states = self.k_proj(hidden_states).view(hidden_shape)
            value_states = self.v_proj(hidden_states).view(hidden_shape) if self.v_proj is not None else key_states

            key_states = self.k_norm(key_states)
            key_states = apply_rotary_pos_emb(key_states, cos, sin, unsqueeze_dim=2)
            key_states = key_states.transpose(1, 2)

            value_states = self.v_norm(value_states)
            value_states = value_states.transpose(1, 2)

        if past_key_values is not None and not self.is_kv_shared_layer:
            key_states, value_states = past_key_values.update(key_states, value_states, self.layer_idx)
        if self.store_full_length_kv:
            shared_kv_states[self.layer_idx] = key_states, value_states

        attn_impl = self.config._attn_implementation
        if attn_impl != "eager" and attn_impl in ALL_ATTENTION_FUNCTIONS:
            attention_interface = ALL_ATTENTION_FUNCTIONS[attn_impl]
        else:
            attention_interface = eager_attention_forward

        attn_output, attn_weights = attention_interface(
            self,
            query_states,
            key_states,
            value_states,
            attention_mask,
            dropout=self.attention_dropout if self.training else 0.0,
            scaling=self.scaling,
            sliding_window=self.sliding_window,
            **kwargs,
        )

        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
        attn_output = self.o_proj(attn_output)
        return attn_output, attn_weights


@use_experts_implementation
class Gemma4TextExperts(nn.Module):
    """Collection of expert weights stored as 3D tensors."""

    def __init__(self, config: Gemma4TextConfig):
        super().__init__()
        self.num_experts = config.num_experts
        self.hidden_dim = config.hidden_size
        self.intermediate_dim = config.moe_intermediate_size
        self.gate_up_proj = nn.Parameter(torch.empty(self.num_experts, 2 * self.intermediate_dim, self.hidden_dim))
        self.down_proj = nn.Parameter(torch.empty(self.num_experts, self.hidden_dim, self.intermediate_dim))
        self.act_fn = ACT2FN[config.hidden_activation]

    def forward(self, hidden_states, top_k_index, top_k_weights):
        final_hidden_states = torch.zeros_like(hidden_states)
        with torch.no_grad():
            expert_mask = torch.nn.functional.one_hot(top_k_index, num_classes=self.num_experts)
            expert_mask = expert_mask.permute(2, 1, 0)
            expert_hit = torch.greater(expert_mask.sum(dim=(-1, -2)), 0).nonzero()

        for expert_idx in expert_hit:
            expert_idx = expert_idx[0]
            if expert_idx == self.num_experts:
                continue
            top_k_pos, token_idx = torch.where(expert_mask[expert_idx])
            current_state = hidden_states[token_idx]
            gate, up = nn.functional.linear(current_state, self.gate_up_proj[expert_idx]).chunk(2, dim=-1)
            current_hidden_states = self.act_fn(gate) * up
            current_hidden_states = nn.functional.linear(current_hidden_states, self.down_proj[expert_idx])
            current_hidden_states = current_hidden_states * top_k_weights[token_idx, top_k_pos, None]
            final_hidden_states.index_add_(0, token_idx, current_hidden_states.to(final_hidden_states.dtype))

        return final_hidden_states


class Gemma4TextRouter(nn.Module):
    def __init__(self, config: Gemma4TextConfig):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.scalar_root_size = self.hidden_size**-0.5
        self.eps = config.rms_norm_eps

        self.norm = Gemma4RMSNorm(self.hidden_size, eps=self.eps, with_scale=False)
        self.proj = nn.Linear(config.hidden_size, config.num_experts, bias=False)
        self.scale = nn.Parameter(torch.ones(self.hidden_size))
        self.per_expert_scale = nn.Parameter(torch.ones(config.num_experts))

    def forward(self, hidden_states):
        hidden_states = self.norm(hidden_states)
        hidden_states = hidden_states * self.scale * self.scalar_root_size

        expert_scores = self.proj(hidden_states)
        router_probabilities = nn.functional.softmax(expert_scores, dim=-1)

        top_k_weights, top_k_index = torch.topk(
            router_probabilities,
            k=self.config.top_k_experts,
            dim=-1,
        )

        top_k_weights /= top_k_weights.sum(dim=-1, keepdim=True)
        top_k_weights = top_k_weights * self.per_expert_scale[top_k_index]

        return router_probabilities, top_k_weights, top_k_index


class Gemma4TextDecoderLayer(GradientCheckpointingLayer):
    def __init__(self, config: Gemma4TextConfig, layer_idx: int):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.layer_idx = layer_idx
        self.self_attn = Gemma4TextAttention(config=config, layer_idx=layer_idx)
        self.mlp = Gemma4TextMLP(config, layer_idx)
        self.input_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
        self.post_attention_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
        self.pre_feedforward_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
        self.post_feedforward_layernorm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
        self.register_buffer("layer_scalar", torch.ones(1))

        self.hidden_size_per_layer_input = config.hidden_size_per_layer_input
        if self.hidden_size_per_layer_input:
            self.act_fn = ACT2FN[config.hidden_activation]
            self.per_layer_input_gate = nn.Linear(self.hidden_size, self.hidden_size_per_layer_input, bias=False)
            self.per_layer_projection = nn.Linear(self.hidden_size_per_layer_input, self.hidden_size, bias=False)
            self.post_per_layer_input_norm = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)

        self.enable_moe_block = config.enable_moe_block
        if self.enable_moe_block:
            self.router = Gemma4TextRouter(config)
            self.experts = Gemma4TextExperts(config)
            self.post_feedforward_layernorm_1 = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
            self.post_feedforward_layernorm_2 = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)
            self.pre_feedforward_layernorm_2 = Gemma4RMSNorm(self.hidden_size, eps=config.rms_norm_eps)

    def forward(
        self,
        hidden_states: torch.Tensor,
        per_layer_input=None,
        shared_kv_states=None,
        position_embeddings=None,
        attention_mask=None,
        position_ids=None,
        past_key_values=None,
        **kwargs,
    ):
        residual = hidden_states

        hidden_states = self.input_layernorm(hidden_states)
        hidden_states, _ = self.self_attn(
            hidden_states=hidden_states,
            position_embeddings=position_embeddings,
            attention_mask=attention_mask,
            shared_kv_states=shared_kv_states,
            position_ids=position_ids,
            past_key_values=past_key_values,
            **kwargs,
        )
        hidden_states = self.post_attention_layernorm(hidden_states)
        hidden_states = residual + hidden_states

        residual = hidden_states
        hidden_states = self.pre_feedforward_layernorm(hidden_states)
        hidden_states = self.mlp(hidden_states)

        if self.enable_moe_block:
            hidden_states_1 = self.post_feedforward_layernorm_1(hidden_states)
            hidden_states_flat = residual.reshape(-1, residual.shape[-1])
            _, top_k_weights, top_k_index = self.router(hidden_states_flat)
            hidden_states_2 = self.pre_feedforward_layernorm_2(hidden_states_flat)
            hidden_states_2 = self.experts(hidden_states_2, top_k_index, top_k_weights)
            hidden_states_2 = hidden_states_2.reshape(residual.shape)
            hidden_states_2 = self.post_feedforward_layernorm_2(hidden_states_2)
            hidden_states = hidden_states_1 + hidden_states_2

        hidden_states = self.post_feedforward_layernorm(hidden_states)
        hidden_states = residual + hidden_states

        if self.hidden_size_per_layer_input:
            residual = hidden_states
            hidden_states = self.per_layer_input_gate(hidden_states)
            hidden_states = self.act_fn(hidden_states)
            hidden_states = hidden_states * per_layer_input
            hidden_states = self.per_layer_projection(hidden_states)
            hidden_states = self.post_per_layer_input_norm(hidden_states)
            hidden_states = residual + hidden_states

        hidden_states *= self.layer_scalar
        return hidden_states


class Gemma4TextScaledWordEmbedding(nn.Embedding):
    """
    This module overrides nn.Embeddings' forward by multiplying with embeddings scale.
    """

    def __init__(self, num_embeddings: int, embedding_dim: int, padding_idx: int, embed_scale: float = 1.0):
        super().__init__(num_embeddings, embedding_dim, padding_idx)
        self.scalar_embed_scale = embed_scale
        self.register_buffer("embed_scale", torch.tensor(embed_scale), persistent=False)

    def forward(self, input_ids: torch.Tensor):
        return super().forward(input_ids) * self.embed_scale.to(self.weight.dtype)


# ---- Model Classes ----

class Gemma4PreTrainedModel(PreTrainedModel):
    config: Gemma4Config
    supports_gradient_checkpointing = True
    _supports_flash_attn = True
    _supports_sdpa = True
    _supports_flex_attn = True
    _can_compile_fullgraph = True
    _supports_attention_backend = True
    _no_split_modules = ["Gemma4TextDecoderLayer", "Gemma4VisionEncoderLayer", "Gemma4AudioLayer"]
    _skip_keys_device_placement = ["past_key_values", "shared_kv_states"]
    input_modalities = ("image", "text", "video", "audio")

    @torch.no_grad()
    def _init_weights(self, module):
        super()._init_weights(module)
        if isinstance(module, Gemma4VisionPatchEmbedder):
            init.ones_(module.position_embedding_table)
        elif isinstance(module, Gemma4AudioRelPositionalEncoding):
            min_timescale = 1.0
            max_timescale = 10000.0
            num_timescales = module.hidden_size // 2
            log_timescale_increment = math.log(max_timescale / min_timescale) / max(num_timescales - 1, 1)
            inv_timescales = min_timescale * torch.exp(torch.arange(num_timescales) * -log_timescale_increment)
            module.inv_timescales.copy_(inv_timescales.unsqueeze(0).unsqueeze(0))
        elif isinstance(module, Gemma4AudioAttention):
            init.constant_(module.softcap, module.attention_logits_soft_cap)
            init.zeros_(module.per_dim_scale)
        elif isinstance(module, Gemma4TextRotaryEmbedding):
            for layer_type, rope_init_fn in module.rope_init_fns.items():
                rope_init_fn_kwargs = {"layer_type": layer_type}
                if layer_type == "full_attention" and module.rope_type[layer_type] == "proportional":
                    rope_init_fn_kwargs["head_dim_key"] = "global_head_dim"

                curr_inv_freq, _ = rope_init_fn(module.config, **rope_init_fn_kwargs)
                getattr(module, f"{layer_type}_inv_freq").copy_(curr_inv_freq)
                getattr(module, f"{layer_type}_original_inv_freq").copy_(curr_inv_freq)
        elif isinstance(module, Gemma4VisionRotaryEmbedding):
            rope_fn = (
                ROPE_INIT_FUNCTIONS[module.rope_type]
                if module.rope_type != "default"
                else module.compute_default_rope_parameters
            )
            buffer_value, _ = rope_fn(module.config)
            module.inv_freq.copy_(buffer_value)
            module.original_inv_freq.copy_(buffer_value)
        elif isinstance(module, Gemma4TextScaledWordEmbedding):
            init.constant_(module.embed_scale, module.scalar_embed_scale)
        elif isinstance(module, Gemma4TextRouter):
            init.ones_(module.scale)
            init.ones_(module.per_expert_scale)
        elif isinstance(module, Gemma4TextExperts):
            std = self.config.initializer_range
            init.normal_(module.gate_up_proj, mean=0.0, std=std)
            init.normal_(module.down_proj, mean=0.0, std=std)
        elif isinstance(module, Gemma4TextDecoderLayer):
            init.ones_(module.layer_scalar)
        elif isinstance(module, Gemma4ClippableLinear) and module.use_clipped_linears:
            init.constant_(module.input_min, -float("inf"))
            init.constant_(module.input_max, float("inf"))
            init.constant_(module.output_min, -float("inf"))
            init.constant_(module.output_max, float("inf"))
        elif isinstance(module, Gemma4VisionModel) and module.config.standardize:
            init.zeros_(module.std_bias)
            init.ones_(module.std_scale)


class Gemma4TextModel(Gemma4PreTrainedModel):
    """The base Gemma 4 language model without a language modeling head."""

    config: Gemma4TextConfig
    input_modalities = ("text",)

    def __init__(self, config: Gemma4TextConfig):
        super().__init__(config)
        self.padding_idx = config.pad_token_id
        self.vocab_size = config.vocab_size

        self.embed_tokens = Gemma4TextScaledWordEmbedding(
            config.vocab_size, config.hidden_size, self.padding_idx, embed_scale=self.config.hidden_size**0.5
        )
        self.layers = nn.ModuleList(
            [Gemma4TextDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
        )
        self.norm = Gemma4RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        self.rotary_emb = Gemma4TextRotaryEmbedding(config)
        self.gradient_checkpointing = False
        self.unique_layer_types = set(self.config.layer_types)

        self.hidden_size_per_layer_input = config.hidden_size_per_layer_input
        if self.hidden_size_per_layer_input:
            self.embed_tokens_per_layer = Gemma4TextScaledWordEmbedding(
                config.vocab_size_per_layer_input,
                config.num_hidden_layers * config.hidden_size_per_layer_input,
                self.padding_idx,
                embed_scale=config.hidden_size_per_layer_input**0.5,
            )
            self.per_layer_input_scale = 2.0**-0.5
            self.per_layer_model_projection = nn.Linear(
                config.hidden_size,
                config.num_hidden_layers * config.hidden_size_per_layer_input,
                bias=False,
            )
            self.per_layer_model_projection_scale = config.hidden_size**-0.5
            self.per_layer_projection_norm = Gemma4RMSNorm(config.hidden_size_per_layer_input, eps=config.rms_norm_eps)

        # Update `_keys_to_ignore_on_load_unexpected` for shared kv layers
        self._keys_to_ignore_on_load_unexpected = []
        for i, layer in enumerate(self.layers):
            if layer.self_attn.is_kv_shared_layer:
                self._keys_to_ignore_on_load_unexpected.extend(
                    [f"layers.{i}.self_attn.{name}" for name in ("k_proj", "v_proj", "k_norm", "v_norm")]
                )

        self.post_init()

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        position_ids=None,
        past_key_values=None,
        inputs_embeds=None,
        per_layer_inputs=None,
        use_cache=None,
        **kwargs,
    ):
        if (input_ids is None) ^ (inputs_embeds is not None):
            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")

        if input_ids is not None:
            inputs_embeds = self.embed_tokens(input_ids)

        if self.hidden_size_per_layer_input:
            if per_layer_inputs is None:
                per_layer_inputs = self.get_per_layer_inputs(input_ids, inputs_embeds)
            per_layer_inputs = self.project_per_layer_inputs(inputs_embeds, per_layer_inputs)

        if use_cache and past_key_values is None:
            past_key_values = DynamicCache()

        if position_ids is None:
            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
            position_ids = torch.arange(inputs_embeds.shape[1], device=inputs_embeds.device) + past_seen_tokens
            position_ids = position_ids.unsqueeze(0)

        # It may already have been prepared by e.g. `generate`
        if not isinstance(causal_mask_mapping := attention_mask, dict):
            mask_kwargs = {
                "config": self.config,
                "input_embeds": inputs_embeds,
                "attention_mask": attention_mask,
                "cache_position": torch.arange(
                    past_key_values.get_seq_length() if past_key_values is not None else 0,
                    (past_key_values.get_seq_length() if past_key_values is not None else 0) + inputs_embeds.shape[1],
                    device=inputs_embeds.device,
                ),
                "past_key_values": past_key_values,
                "position_ids": position_ids,
            }
            causal_mask_mapping = {
                "full_attention": create_causal_mask(**mask_kwargs),
                "sliding_attention": create_sliding_window_causal_mask(**mask_kwargs),
            }

        hidden_states = inputs_embeds
        position_embeddings = {}
        for layer_type in self.unique_layer_types:
            position_embeddings[layer_type] = self.rotary_emb(hidden_states, position_ids, layer_type)

        shared_kv_states = {}

        for i, decoder_layer in enumerate(self.layers[: self.config.num_hidden_layers]):
            per_layer_input = per_layer_inputs[:, :, i, :] if per_layer_inputs is not None else None

            hidden_states = decoder_layer(
                hidden_states,
                per_layer_input,
                shared_kv_states=shared_kv_states,
                position_embeddings=position_embeddings[self.config.layer_types[i]],
                attention_mask=causal_mask_mapping[self.config.layer_types[i]],
                position_ids=position_ids,
                past_key_values=past_key_values,
                **kwargs,
            )

        hidden_states = self.norm(hidden_states)

        return BaseModelOutputWithPast(
            last_hidden_state=hidden_states,
            past_key_values=past_key_values,
        )

    def get_per_layer_inputs(self, input_ids, inputs_embeds):
        if not self.hidden_size_per_layer_input:
            raise RuntimeError("Model not configured with per-layer embeddings.")

        if input_ids is None:
            with torch.no_grad():
                input_ids = (
                    (
                        inputs_embeds[:, :, None, :]
                        == self.embed_tokens.weight[None, None, :, :] * self.config.hidden_size**0.5
                    )
                    .all(dim=3)
                    .nonzero()[:, 2]
                )
                try:
                    input_ids = input_ids.view(inputs_embeds.shape[:2])
                except RuntimeError:
                    raise RuntimeError(
                        "Cannot reverse embedding to recover input_ids. Provide both `input_ids` and `inputs_embeds`."
                    )

        return self.embed_tokens_per_layer(input_ids).reshape(
            *input_ids.shape,
            self.config.num_hidden_layers,
            self.hidden_size_per_layer_input,
        )

    def project_per_layer_inputs(self, inputs_embeds, per_layer_inputs=None):
        if not self.hidden_size_per_layer_input:
            raise RuntimeError("Model not configured with per-layer embeddings.")

        per_layer_projection = self.per_layer_model_projection(inputs_embeds) * self.per_layer_model_projection_scale
        per_layer_projection = per_layer_projection.reshape(
            *inputs_embeds.shape[:-1],
            self.config.num_hidden_layers,
            self.hidden_size_per_layer_input,
        )
        per_layer_projection = self.per_layer_projection_norm(per_layer_projection)

        if per_layer_inputs is None:
            return per_layer_projection

        return (per_layer_projection + per_layer_inputs) * self.per_layer_input_scale


class Gemma4ForCausalLM(Gemma4PreTrainedModel, GenerationMixin):
    """The base Gemma 4 language model with a language modeling head."""

    _tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"}
    config: Gemma4TextConfig
    base_model_prefix = "model"

    def __init__(self, config: Gemma4TextConfig):
        super().__init__(config)
        self.model = Gemma4TextModel(config)
        self.vocab_size = config.vocab_size
        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
        self._keys_to_ignore_on_load_unexpected = [
            f"model.{name}" for name in self.model._keys_to_ignore_on_load_unexpected
        ]
        self.post_init()

    @can_return_tuple
    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        position_ids=None,
        past_key_values=None,
        inputs_embeds=None,
        labels=None,
        use_cache=None,
        logits_to_keep=0,
        **kwargs,
    ):
        outputs = self.model(
            input_ids=input_ids,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            use_cache=use_cache,
            **kwargs,
        )

        hidden_states = outputs.last_hidden_state
        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
        logits = self.lm_head(hidden_states[:, slice_indices, :])
        if self.config.final_logit_softcapping is not None:
            logits = logits / self.config.final_logit_softcapping
            logits = torch.tanh(logits)
            logits = logits * self.config.final_logit_softcapping

        loss = None
        if labels is not None:
            loss = self.loss_function(logits, labels, self.vocab_size, **kwargs)

        return CausalLMOutputWithPast(
            loss=loss,
            logits=logits,
            past_key_values=outputs.past_key_values,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )


def sliding_window_mask_function(sliding_window):
    """Creates uni/bidirectional attention mask with sliding window."""

    def inner_mask(batch_idx, head_idx, q_idx, kv_idx):
        left_window_size, right_window_size = sliding_window

        dist = q_idx - kv_idx
        left_mask = (dist >= 0) & (dist < left_window_size)
        right_mask = (dist < 0) & (-dist < right_window_size)
        return left_mask | right_mask

    return inner_mask


class Gemma4AudioModel(Gemma4PreTrainedModel):
    """An audio encoder based on the Universal Speech Model architecture."""

    config: Gemma4AudioConfig
    main_input_name = "input_features"
    base_model_prefix = "model.audio_tower"

    def __init__(self, config: Gemma4AudioConfig):
        super().__init__(config)
        self.config = config

        self.subsample_conv_projection = Gemma4AudioSubSampleConvProjection(config)
        self.rel_pos_enc = Gemma4AudioRelPositionalEncoding(config)
        self.layers = nn.ModuleList(
            [Gemma4AudioLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
        )
        self.output_proj = nn.Linear(config.hidden_size, config.output_proj_dims, bias=True)

        self.post_init()

    def _convert_4d_mask_to_blocked_5d(self, mask_4d: torch.Tensor) -> torch.Tensor:
        batch_size, _, seq_len, _ = mask_4d.shape
        device = mask_4d.device

        chunk_size = self.config.attention_chunk_size
        max_past_horizon = self.config.attention_context_left - 1
        max_future_horizon = self.config.attention_context_right

        num_blocks = (seq_len + chunk_size - 1) // chunk_size
        padded_seq_len = num_blocks * chunk_size
        pad_amount = padded_seq_len - seq_len

        mask_4d = F.pad(mask_4d, (0, pad_amount, 0, pad_amount), value=False)
        mask_5d = mask_4d.reshape(batch_size, 1, num_blocks, chunk_size, padded_seq_len)
        mask_5d = F.pad(mask_5d, (max_past_horizon, max_future_horizon), value=False)

        block_starts = torch.arange(num_blocks, device=device) * chunk_size
        offsets = torch.arange(chunk_size + max_past_horizon + max_future_horizon, device=device)
        kv_indices = block_starts[:, None] + offsets[None, :]
        kv_indices = kv_indices[None, None, :, None, :].expand(batch_size, 1, -1, chunk_size, -1)

        return mask_5d.gather(-1, kv_indices)

    def forward(self, input_features, attention_mask=None, **kwargs):
        hidden_states, output_mask = self.subsample_conv_projection(input_features, attention_mask)
        position_embeddings = self.rel_pos_enc(hidden_states)

        attention_mask_4d = create_bidirectional_mask(
            config=self.config,
            inputs_embeds=hidden_states,
            attention_mask=output_mask,
        )
        attention_mask_4d = self._convert_4d_mask_to_blocked_5d(attention_mask_4d)

        for encoder_layer in self.layers[: self.config.num_hidden_layers]:
            hidden_states = encoder_layer(
                hidden_states,
                attention_mask=attention_mask_4d,
                position_embeddings=position_embeddings,
                **kwargs,
            )

        hidden_states = self.output_proj(hidden_states)
        return Gemma4AudioModelOutput(last_hidden_state=hidden_states, attention_mask=output_mask)


class Gemma4VisionModel(Gemma4PreTrainedModel):
    """The Gemma 4 Vision Encoder."""

    config = Gemma4VisionConfig

    def __init__(self, config: Gemma4VisionConfig):
        super().__init__(config)
        self.patch_embedder = Gemma4VisionPatchEmbedder(config)
        self.encoder = Gemma4VisionEncoder(config)
        self.pooler = Gemma4VisionPooler(config)

        if self.config.standardize:
            self.register_buffer("std_bias", torch.empty(self.config.hidden_size))
            self.register_buffer("std_scale", torch.empty(self.config.hidden_size))

        self.post_init()

    def forward(self, pixel_values, pixel_position_ids, **kwargs):
        pooling_kernel_size = self.config.pooling_kernel_size
        output_length = pixel_values.shape[-2] // (pooling_kernel_size * pooling_kernel_size)

        padding_positions = (pixel_position_ids == -1).all(dim=-1)
        inputs_embeds = self.patch_embedder(pixel_values, pixel_position_ids, padding_positions)
        output = self.encoder(
            inputs_embeds=inputs_embeds,
            attention_mask=~padding_positions,
            pixel_position_ids=pixel_position_ids,
            **kwargs,
        )

        hidden_states, pooler_mask = self.pooler(
            hidden_states=output.last_hidden_state,
            pixel_position_ids=pixel_position_ids,
            padding_positions=padding_positions,
            output_length=output_length,
        )

        hidden_states = hidden_states[pooler_mask]

        if self.config.standardize:
            hidden_states = (hidden_states - self.std_bias) * self.std_scale

        return BaseModelOutputWithPast(last_hidden_state=hidden_states)


class Gemma4MultimodalEmbedder(nn.Module):
    """Embeds token ids or soft tokens for multimodal content into language model space."""

    def __init__(self, multimodal_config, text_config):
        super().__init__()

        self.multimodal_hidden_size = getattr(multimodal_config, "output_proj_dims", multimodal_config.hidden_size)
        self.eps = multimodal_config.rms_norm_eps
        self.text_hidden_size = text_config.hidden_size
        self.embedding_projection = nn.Linear(self.multimodal_hidden_size, self.text_hidden_size, bias=False)
        self.embedding_pre_projection_norm = Gemma4RMSNorm(self.multimodal_hidden_size, eps=self.eps, with_scale=False)

    def forward(self, inputs_embeds: torch.Tensor) -> torch.Tensor:
        embs_normed = self.embedding_pre_projection_norm(inputs_embeds)
        return self.embedding_projection(embs_normed)


def token_type_ids_mask_function(token_type_ids, image_group_ids):
    if token_type_ids is None:
        return None

    def inner_mask(batch_idx, head_idx, q_idx, kv_idx):
        seq_length = image_group_ids.shape[-1]

        q_idx_clamped = q_idx.clamp(max=seq_length - 1)
        kv_idx_clamped = kv_idx.clamp(max=seq_length - 1)

        q_group = image_group_ids[batch_idx, q_idx_clamped]
        kv_group = image_group_ids[batch_idx, kv_idx_clamped]
        q_group = torch.where(q_idx < seq_length, q_group, -1)
        kv_group = torch.where(kv_idx < seq_length, kv_group, -1)
        return (q_group == kv_group) & (q_group >= 0)

    return inner_mask


def create_causal_mask_mapping(
    config,
    inputs_embeds,
    attention_mask,
    past_key_values,
    position_ids,
    mm_token_type_ids=None,
    pixel_values=None,
    is_training=False,
    is_first_iteration=None,
    **kwargs,
):
    if is_training and mm_token_type_ids is None:
        raise ValueError("`mm_token_type_ids` is required as a model input when training")

    past_seq_len = past_key_values.get_seq_length() if past_key_values is not None else 0
    cur_seq_len = inputs_embeds.shape[1]
    cache_position = torch.arange(past_seq_len, past_seq_len + cur_seq_len, device=inputs_embeds.device)

    mask_kwargs = {
        "config": config.get_text_config(),
        "input_embeds": inputs_embeds,
        "attention_mask": attention_mask,
        "cache_position": cache_position,
        "past_key_values": past_key_values,
        "position_ids": position_ids,
    }
    sliding_mask_kwargs = mask_kwargs.copy()

    is_first_iteration = (
        is_first_iteration
        if is_first_iteration is not None
        else (past_key_values is None or pixel_values is not None)
    )
    if mm_token_type_ids is not None and is_first_iteration:
        is_vision = (mm_token_type_ids == 1) | (mm_token_type_ids == 2)
        is_prev_vision = torch.roll(is_vision, shifts=1, dims=-1)
        is_prev_vision[..., 0] = False
        new_vision_starts = is_vision & ~is_prev_vision
        vision_group_ids = torch.cumsum(new_vision_starts.int(), dim=1) - 1
        vision_group_ids = torch.where(is_vision, vision_group_ids, -1)
        sliding_mask_kwargs["or_mask_function"] = token_type_ids_mask_function(
            mm_token_type_ids.to(inputs_embeds.device), vision_group_ids
        )

    return {
        "full_attention": create_causal_mask(**mask_kwargs),
        "sliding_attention": create_sliding_window_causal_mask(**sliding_mask_kwargs),
    }


class Gemma4Model(Gemma4PreTrainedModel):
    """
    The base Gemma 4 model comprising a vision backbone, an audio backbone,
    and a language model without a language modeling head.
    """

    accepts_loss_kwargs = False

    def __init__(self, config: Gemma4Config):
        super().__init__(config)
        self.vocab_size = config.text_config.vocab_size

        self.language_model = Gemma4TextModel(config.text_config)
        self.vocab_size_per_layer_input = config.text_config.vocab_size_per_layer_input
        self.vision_tower = Gemma4VisionModel(config.vision_config) if config.vision_config is not None else None
        self.embed_vision = (
            Gemma4MultimodalEmbedder(config.vision_config, config.text_config)
            if config.vision_config is not None
            else None
        )
        self.audio_tower = Gemma4AudioModel(config.audio_config) if config.audio_config is not None else None
        self.embed_audio = (
            Gemma4MultimodalEmbedder(config.audio_config, config.text_config)
            if config.audio_config is not None
            else None
        )
        self._keys_to_ignore_on_load_unexpected = [
            f"language_model.{name}" for name in self.language_model._keys_to_ignore_on_load_unexpected
        ]
        self.post_init()

    def get_input_embeddings(self):
        return self.language_model.get_input_embeddings()

    def set_input_embeddings(self, value):
        self.language_model.set_input_embeddings(value)

    @can_return_tuple
    def get_image_features(self, pixel_values, image_position_ids=None, **kwargs):
        vision_outputs = self.vision_tower(
            pixel_values=pixel_values,
            pixel_position_ids=image_position_ids,
            **kwargs,
        )
        last_hidden_state = vision_outputs.last_hidden_state
        vision_outputs.pooler_output = self.embed_vision(inputs_embeds=last_hidden_state)
        return vision_outputs

    def get_placeholder_mask(self, input_ids=None, inputs_embeds=None):
        if input_ids is not None:
            special_image_mask = input_ids == self.config.image_token_id
            special_video_mask = input_ids == self.config.video_token_id
            special_audio_mask = input_ids == self.config.audio_token_id
        else:
            special_image_mask = (
                inputs_embeds
                == self.get_input_embeddings()(
                    torch.tensor(self.config.image_token_id, dtype=torch.long, device=inputs_embeds.device)
                )
            ).all(-1)
            special_video_mask = (
                inputs_embeds
                == self.get_input_embeddings()(
                    torch.tensor(self.config.video_token_id, dtype=torch.long, device=inputs_embeds.device)
                )
            ).all(-1)
            special_audio_mask = (
                inputs_embeds
                == self.get_input_embeddings()(
                    torch.tensor(self.config.audio_token_id, dtype=torch.long, device=inputs_embeds.device)
                )
            ).all(-1)

        return special_image_mask, special_video_mask, special_audio_mask

    @can_return_tuple
    def forward(
        self,
        input_ids=None,
        pixel_values=None,
        pixel_values_videos=None,
        input_features=None,
        attention_mask=None,
        input_features_mask=None,
        position_ids=None,
        past_key_values=None,
        mm_token_type_ids=None,
        inputs_embeds=None,
        use_cache=None,
        image_position_ids=None,
        video_position_ids=None,
        **kwargs,
    ):
        if (input_ids is None) ^ (inputs_embeds is not None):
            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")

        image_mask, video_mask, audio_mask = self.get_placeholder_mask(input_ids, inputs_embeds)
        multimodal_mask = image_mask | video_mask | audio_mask

        llm_input_ids = None
        if inputs_embeds is None:
            llm_input_ids = input_ids.clone()
            llm_input_ids[multimodal_mask] = self.config.text_config.pad_token_id
            inputs_embeds = self.get_input_embeddings()(llm_input_ids)

        if self.config.get_text_config().hidden_size_per_layer_input:
            pad_embedding = self.language_model.embed_tokens.weight[self.config.text_config.pad_token_id, :]
            llm_inputs_embeds = torch.where(multimodal_mask[..., None], pad_embedding.view(1, 1, -1), inputs_embeds)
            per_layer_inputs = self.language_model.get_per_layer_inputs(llm_input_ids, llm_inputs_embeds)
        else:
            per_layer_inputs = None

        if pixel_values is not None:
            image_features = self.get_image_features(pixel_values, image_position_ids, return_dict=True).pooler_output
            image_features = image_features.to(inputs_embeds.device, inputs_embeds.dtype)

            n_image_tokens = image_mask.sum()
            image_mask_exp = image_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
            if inputs_embeds[image_mask_exp].numel() != image_features.numel():
                raise ValueError(
                    f"Image features and image tokens do not match, tokens: {n_image_tokens}, features:"
                    f" {image_features.shape[0]}"
                )
            inputs_embeds = inputs_embeds.masked_scatter(image_mask_exp, image_features.to(inputs_embeds.device))

        if pixel_values_videos is not None:
            video_features = self.get_video_features(pixel_values_videos, video_position_ids, return_dict=True).pooler_output
            video_features = video_features.to(inputs_embeds.device, inputs_embeds.dtype)

            n_video_tokens = video_mask.sum()
            video_mask_exp = video_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
            if inputs_embeds[video_mask_exp].numel() != video_features.numel():
                raise ValueError(
                    f"Video features and video tokens do not match, tokens: {n_video_tokens}, features:"
                    f" {video_features.shape[0]}"
                )
            inputs_embeds = inputs_embeds.masked_scatter(video_mask_exp, video_features.to(inputs_embeds.device))

        if input_features is not None and input_features_mask is not None:
            audio_output = self.get_audio_features(input_features, input_features_mask, return_dict=True)
            audio_features = audio_output.pooler_output
            audio_mask_from_encoder = audio_output.attention_mask
            audio_features = audio_features[audio_mask_from_encoder]

            n_audio_tokens = audio_mask.sum()
            audio_mask_exp = audio_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
            if inputs_embeds[audio_mask_exp].numel() != audio_features.numel():
                raise ValueError(
                    f"Audio features and audio tokens do not match, tokens: {n_audio_tokens}, features:"
                    f" {audio_features.shape[0] * audio_features.shape[1]}"
                )
            inputs_embeds = inputs_embeds.masked_scatter(audio_mask_exp, audio_features.to(inputs_embeds.device))

        if position_ids is None:
            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
            position_ids = torch.arange(inputs_embeds.shape[1], device=inputs_embeds.device) + past_seen_tokens
            position_ids = position_ids.unsqueeze(0)

        if not isinstance(causal_mask_mapping := attention_mask, dict):
            if self.config.get_text_config().use_bidirectional_attention == "vision":
                causal_mask_mapping = create_causal_mask_mapping(
                    self.config,
                    inputs_embeds,
                    attention_mask,
                    past_key_values,
                    position_ids,
                    mm_token_type_ids,
                    pixel_values,
                    is_training=self.training,
                )
            else:
                past_seq_len = past_key_values.get_seq_length() if past_key_values is not None else 0
                cur_seq_len = inputs_embeds.shape[1]
                cache_position = torch.arange(past_seq_len, past_seq_len + cur_seq_len, device=inputs_embeds.device)
                causal_mask_mapping = create_masks_for_generate(
                    self.config,
                    inputs_embeds,
                    attention_mask,
                    cache_position,
                    past_key_values,
                    position_ids,
                )

        outputs = self.language_model(
            per_layer_inputs=per_layer_inputs,
            attention_mask=causal_mask_mapping,
            position_ids=position_ids,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            use_cache=use_cache,
            return_dict=True,
            **kwargs,
        )

        image_features_out = image_features if pixel_values is not None else None
        audio_features_out = audio_features if input_features is not None else None

        return Gemma4ModelOutputWithPast(
            last_hidden_state=outputs.last_hidden_state,
            past_key_values=outputs.past_key_values,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
            image_hidden_states=image_features_out,
            audio_hidden_states=audio_features_out,
        )

    @can_return_tuple
    def get_audio_features(self, input_features, input_features_mask, **kwargs):
        if self.audio_tower is None:
            raise ValueError(
                "Audio features were requested, but the model was initialized without an audio_config."
            )

        audio_outputs = self.audio_tower(input_features, input_features_mask, return_dict=True, **kwargs)
        audio_outputs.pooler_output = self.embed_audio(inputs_embeds=audio_outputs.last_hidden_state)

        return audio_outputs

    @can_return_tuple
    def get_video_features(self, pixel_values_videos, video_position_ids=None, **kwargs):
        pixel_values_videos = pixel_values_videos.flatten(0, 1)
        video_position_ids = video_position_ids.flatten(0, 1)
        vision_outputs = self.vision_tower(
            pixel_values=pixel_values_videos,
            pixel_position_ids=video_position_ids,
            **kwargs,
        )
        last_hidden_state = vision_outputs.last_hidden_state
        vision_outputs.pooler_output = self.embed_vision(inputs_embeds=last_hidden_state)
        return vision_outputs


class Gemma4ForConditionalGeneration(Gemma4PreTrainedModel, GenerationMixin):
    """
    The base Gemma 4 model comprising a vision backbone, an audio backbone,
    a language model, and a language modeling head.
    """

    _tied_weights_keys = {"lm_head.weight": "model.language_model.embed_tokens.weight"}
    base_model_prefix = "model"

    def __init__(self, config: Gemma4Config):
        super().__init__(config)
        self.model = Gemma4Model(config)
        self.lm_head = nn.Linear(config.text_config.hidden_size, config.text_config.vocab_size, bias=False)
        self._keys_to_ignore_on_load_unexpected = [
            f"model.{name}" for name in self.model._keys_to_ignore_on_load_unexpected
        ]
        self.post_init()

    def get_input_embeddings(self):
        return self.model.get_input_embeddings()

    def set_input_embeddings(self, value):
        self.model.set_input_embeddings(value)

    def get_image_features(self, pixel_values, image_position_ids=None, **kwargs):
        return self.model.get_image_features(pixel_values, image_position_ids, **kwargs)

    @can_return_tuple
    def forward(
        self,
        input_ids=None,
        pixel_values=None,
        pixel_values_videos=None,
        input_features=None,
        attention_mask=None,
        input_features_mask=None,
        position_ids=None,
        image_position_ids=None,
        video_position_ids=None,
        past_key_values=None,
        mm_token_type_ids=None,
        inputs_embeds=None,
        labels=None,
        use_cache=None,
        logits_to_keep=0,
        **kwargs,
    ):
        outputs = self.model(
            input_ids=input_ids,
            pixel_values=pixel_values,
            pixel_values_videos=pixel_values_videos,
            input_features=input_features,
            attention_mask=attention_mask,
            input_features_mask=input_features_mask,
            position_ids=position_ids,
            past_key_values=past_key_values,
            mm_token_type_ids=mm_token_type_ids,
            inputs_embeds=inputs_embeds,
            use_cache=use_cache,
            image_position_ids=image_position_ids,
            video_position_ids=video_position_ids,
            return_dict=True,
            **kwargs,
        )

        hidden_states = outputs.last_hidden_state
        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
        logits = self.lm_head(hidden_states[:, slice_indices, :])
        if (final_logit_softcapping := self.config.get_text_config().final_logit_softcapping) is not None:
            logits = logits / final_logit_softcapping
            logits = torch.tanh(logits)
            logits = logits * final_logit_softcapping

        loss = None
        if labels is not None:
            logits = logits.float()
            shift_logits = logits[..., :-1, :]
            shift_labels = labels[..., 1:]
            if attention_mask is not None:
                shift_attention_mask = attention_mask[:, -shift_logits.shape[1]:].to(logits.device)
                shift_logits = shift_logits[shift_attention_mask.to(logits.device) != 0].contiguous()
                shift_labels = shift_labels[shift_attention_mask.to(shift_labels.device) != 0].contiguous()
            else:
                shift_logits = shift_logits.contiguous()
                shift_labels = shift_labels.contiguous()
            loss_fct = nn.CrossEntropyLoss()
            flat_logits = shift_logits.view(-1, self.config.get_text_config().vocab_size)
            flat_labels = shift_labels.view(-1).to(shift_logits.device)
            loss = loss_fct(flat_logits, flat_labels)

        return Gemma4CausalLMOutputWithPast(
            loss=loss,
            logits=logits,
            past_key_values=outputs.past_key_values,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
            image_hidden_states=outputs.image_hidden_states,
            audio_hidden_states=outputs.audio_hidden_states,
        )

    def prepare_inputs_for_generation(
        self,
        input_ids,
        past_key_values=None,
        inputs_embeds=None,
        position_ids=None,
        pixel_values=None,
        pixel_values_videos=None,
        input_features=None,
        attention_mask=None,
        input_features_mask=None,
        token_type_ids=None,
        use_cache=True,
        logits_to_keep=None,
        labels=None,
        is_first_iteration=False,
        **kwargs,
    ):
        model_inputs = super().prepare_inputs_for_generation(
            input_ids,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            attention_mask=attention_mask,
            position_ids=position_ids,
            use_cache=use_cache,
            logits_to_keep=logits_to_keep,
            token_type_ids=token_type_ids,
            is_first_iteration=is_first_iteration,
            **kwargs,
        )

        if is_first_iteration or not use_cache:
            model_inputs["pixel_values"] = pixel_values
            model_inputs["pixel_values_videos"] = pixel_values_videos
            model_inputs["input_features"] = input_features
            model_inputs["input_features_mask"] = input_features_mask

        return model_inputs

    @staticmethod
    def create_masks_for_generate(
        config,
        inputs_embeds,
        attention_mask,
        past_key_values,
        position_ids,
        mm_token_type_ids=None,
        is_first_iteration=False,
        **kwargs,
    ):
        if getattr(config.get_text_config(), "use_bidirectional_attention", None) == "vision":
            return create_causal_mask_mapping(
                config,
                inputs_embeds,
                attention_mask,
                past_key_values,
                position_ids,
                mm_token_type_ids,
                is_first_iteration=is_first_iteration,
                **{k: v for k, v in kwargs.items() if k != "pixel_values"},
            )
        else:
            past_seq_len = past_key_values.get_seq_length() if past_key_values is not None else 0
            cur_seq_len = inputs_embeds.shape[1]
            cache_position = torch.arange(past_seq_len, past_seq_len + cur_seq_len, device=inputs_embeds.device)
            return create_masks_for_generate(
                config, inputs_embeds, attention_mask, cache_position, past_key_values, position_ids, **kwargs
            )


__all__ = [
    "Gemma4AudioModel",
    "Gemma4ForCausalLM",
    "Gemma4ForConditionalGeneration",
    "Gemma4Model",
    "Gemma4PreTrainedModel",
    "Gemma4TextModel",
    "Gemma4VisionModel",
]